rsync - Jul 2003 - rsync 2.5.6 globbing bug

$ rsync -auv \
ftp.funet.fi::CPAN/modules/by-category/05_Networking_Devices_IPC/Net/* . \
| grep -v MOTD:

rsync: connection unexpectedly closed (1598 bytes read so far)
rsync error: error in rsync protocol data stream (code 12) at io.c(165)

This is because the array used to hold the expanded filename glob (argv in 
rsync_module(), clientserver.c) is only 1000 entries in size, and the above 
archive directory contains more than 1000 files.  argv should probably be 
allocated with malloc and grown as necessary with realloc.  I'm happy to 
submit a patch if that is acceptable.

Please reply direct as I'm not ton the mailing list,

Thanks,

-- 
Alan Burlison
--

On Wed, Jul 30, 2003 at 10:44:37PM +0100, Alan Burlison
wrote:
> $ rsync -auv \
> ftp.funet.fi::CPAN/modules/by-category/05_Networking_Devices_IPC/Net/* . \
> | grep -v MOTD:
Cockpit error.
This isn't DOS.  Don't use * when you want the all directory contents.
	rsync -auv
ftp.funet.fi::CPAN/modules/by-category/05_Networking_Devices_IPC/Net/ .
Will work correctly and is safe from the local shell
globbing which would have given you an error message had you
been using csh with default options.
> rsync: connection unexpectedly closed (1598 bytes read so far)
> rsync error: error in rsync protocol data stream (code 12) at io.c(165)
> 
> This is because the array used to hold the expanded filename glob (argv in 
> rsync_module(), clientserver.c) is only 1000 entries in size, and the above
> archive directory contains more than 1000 files.  argv should probably be 
> allocated with malloc and grown as necessary with realloc.  I'm happy
to
> submit a patch if that is acceptable.
> 
> Please reply direct as I'm not ton the mailing list,
-- 
________________________________________________________________
	J.W. Schultz            Pegasystems Technologies
	email address:		jw@pegasys.ws

		Remember Cernan and Schmitt

jw schultz wrote:
>>I do entirely agree with your point about sanity over performance, and I
>>think the current mysterious failure mode is not at all helpful.
> 
> I just looked and here i do agree with you.  This is
> apparently just silently failing when it should kick out an
> error message to the client.  "Argument list too long" would
> be appropriate.
Fixing it would be even nicer :-)

-- 
Alan Burlison
--